Date
AI Overview
Why it matters
Traditional language models have limited adaptability and reasoning. Synthesizing new agentic paradigms through an educational lens helps clarify their pedagogical affordances, potential benefits, and challenges in learning environments.
Study at a glance
- Question: How can agentic AI paradigms support education, and does a multi-agent system improve automated essay scoring accuracy?
- Setting and data: ASAP 2.0 Dataset comprising about 17,000 student-written argumentative essays scored on a 1 to 6 rubric; literature search of 93 studies.
- Method: Systematic literature review following PRISMA guidelines and benchmark evaluation comparing a prototype multi-agent scoring system to stand-alone language models.
- AI tools: Multi-Agent Scoring System (MASS), GPT-4o, DeepSeek 67B, DeepSeek 1.3B, Llama 3.3 70B
Key findings
- The multi-agent scoring system achieved a lower mean absolute error (0.561) than stand-alone GPT-4o (0.613), Llama 3.3 70B (0.614), and DeepSeek models.
- Llama 3.3 70B achieved a lower standard deviation of error (0.783) than the multi-agent system (0.830), indicating tighter clustering of predictions.
- The supervisor agent requested additional reports in roughly 14% of essays where subagent content and language scores differed by more than two rubric points.
Summary drafted with AI from the full paper and may contain errors. See the full paper for details.
The primary goal of this study is to analyze agentic workflows in education according to the proposed four major technological paradigms: reflection, planning, tool use, and multi-agent collaboration. We critically examine the role of AI agents in education through these key design paradigms, exploring their advantages, applications, and challenges. Second, to illustrate the practical potential of agentic systems, we present a proof-of-concept application: a multi-agent framework for automated essay scoring. Preliminary results suggest this agentic approach may offer improved consistency compared to stand-alone LLMs. Our findings highlight the transformative potential of AI agents in educational settings while underscoring the need for further research into their interpretability and trustworthiness.
Research Categories
What is the application?
Who age?
Why use AI?
AI tools
Datasets
Publication Source
Publisher
arXiv
